git.oblomov.eu Git - git/blob - builtin/index-pack.c

   1 #include "builtin.h"
   2 #include "delta.h"
   3 #include "pack.h"
   4 #include "csum-file.h"
   5 #include "blob.h"
   6 #include "commit.h"
   7 #include "tag.h"
   8 #include "tree.h"
   9 #include "progress.h"
  10 #include "fsck.h"
  11 #include "exec_cmd.h"
  12 #include "streaming.h"
  13 #include "thread-utils.h"
  14
  15 static const char index_pack_usage[] =
  16 "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
  17
  18 struct object_entry {
  19         struct pack_idx_entry idx;
  20         unsigned long size;
  21         unsigned int hdr_size;
  22         enum object_type type;
  23         enum object_type real_type;
  24         unsigned delta_depth;
  25         int base_object_no;
  26 };
  27
  28 union delta_base {
  29         unsigned char sha1[20];
  30         off_t offset;
  31 };
  32
  33 struct base_data {
  34         struct base_data *base;
  35         struct base_data *child;
  36         struct object_entry *obj;
  37         void *data;
  38         unsigned long size;
  39         int ref_first, ref_last;
  40         int ofs_first, ofs_last;
  41 };
  42
  43 struct thread_local {
  44 #ifndef NO_PTHREADS
  45         pthread_t thread;
  46 #endif
  47         struct base_data *base_cache;
  48         size_t base_cache_used;
  49         int pack_fd;
  50 };
  51
  52 /*
  53  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  54  * to memcmp() only the first 20 bytes.
  55  */
  56 #define UNION_BASE_SZ   20
  57
  58 #define FLAG_LINK (1u<<20)
  59 #define FLAG_CHECKED (1u<<21)
  60
  61 struct delta_entry {
  62         union delta_base base;
  63         int obj_no;
  64 };
  65
  66 static struct object_entry *objects;
  67 static struct delta_entry *deltas;
  68 static struct thread_local nothread_data;
  69 static int nr_objects;
  70 static int nr_deltas;
  71 static int nr_resolved_deltas;
  72 static int nr_threads;
  73
  74 static int from_stdin;
  75 static int strict;
  76 static int do_fsck_object;
  77 static int verbose;
  78 static int show_stat;
  79 static int check_self_contained_and_connected;
  80
  81 static struct progress *progress;
  82
  83 /* We always read in 4kB chunks. */
  84 static unsigned char input_buffer[4096];
  85 static unsigned int input_offset, input_len;
  86 static off_t consumed_bytes;
  87 static unsigned deepest_delta;
  88 static git_SHA_CTX input_ctx;
  89 static uint32_t input_crc32;
  90 static int input_fd, output_fd;
  91 static const char *curr_pack;
  92
  93 #ifndef NO_PTHREADS
  94
  95 static struct thread_local *thread_data;
  96 static int nr_dispatched;
  97 static int threads_active;
  98
  99 static pthread_mutex_t read_mutex;
 100 #define read_lock()             lock_mutex(&read_mutex)
 101 #define read_unlock()           unlock_mutex(&read_mutex)
 102
 103 static pthread_mutex_t counter_mutex;
 104 #define counter_lock()          lock_mutex(&counter_mutex)
 105 #define counter_unlock()        unlock_mutex(&counter_mutex)
 106
 107 static pthread_mutex_t work_mutex;
 108 #define work_lock()             lock_mutex(&work_mutex)
 109 #define work_unlock()           unlock_mutex(&work_mutex)
 110
 111 static pthread_mutex_t deepest_delta_mutex;
 112 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 113 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 114
 115 static pthread_mutex_t type_cas_mutex;
 116 #define type_cas_lock()         lock_mutex(&type_cas_mutex)
 117 #define type_cas_unlock()       unlock_mutex(&type_cas_mutex)
 118
 119 static pthread_key_t key;
 120
 121 static inline void lock_mutex(pthread_mutex_t *mutex)
 122 {
 123         if (threads_active)
 124                 pthread_mutex_lock(mutex);
 125 }
 126
 127 static inline void unlock_mutex(pthread_mutex_t *mutex)
 128 {
 129         if (threads_active)
 130                 pthread_mutex_unlock(mutex);
 131 }
 132
 133 /*
 134  * Mutex and conditional variable can't be statically-initialized on Windows.
 135  */
 136 static void init_thread(void)
 137 {
 138         int i;
 139         init_recursive_mutex(&read_mutex);
 140         pthread_mutex_init(&counter_mutex, NULL);
 141         pthread_mutex_init(&work_mutex, NULL);
 142         pthread_mutex_init(&type_cas_mutex, NULL);
 143         if (show_stat)
 144                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 145         pthread_key_create(&key, NULL);
 146         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 147         for (i = 0; i < nr_threads; i++) {
 148                 thread_data[i].pack_fd = open(curr_pack, O_RDONLY);
 149                 if (thread_data[i].pack_fd == -1)
 150                         die_errno(_("unable to open %s"), curr_pack);
 151         }
 152
 153         threads_active = 1;
 154 }
 155
 156 static void cleanup_thread(void)
 157 {
 158         int i;
 159         if (!threads_active)
 160                 return;
 161         threads_active = 0;
 162         pthread_mutex_destroy(&read_mutex);
 163         pthread_mutex_destroy(&counter_mutex);
 164         pthread_mutex_destroy(&work_mutex);
 165         pthread_mutex_destroy(&type_cas_mutex);
 166         if (show_stat)
 167                 pthread_mutex_destroy(&deepest_delta_mutex);
 168         for (i = 0; i < nr_threads; i++)
 169                 close(thread_data[i].pack_fd);
 170         pthread_key_delete(key);
 171         free(thread_data);
 172 }
 173
 174 #else
 175
 176 #define read_lock()
 177 #define read_unlock()
 178
 179 #define counter_lock()
 180 #define counter_unlock()
 181
 182 #define work_lock()
 183 #define work_unlock()
 184
 185 #define deepest_delta_lock()
 186 #define deepest_delta_unlock()
 187
 188 #endif
 189
 190
 191 static int mark_link(struct object *obj, int type, void *data)
 192 {
 193         if (!obj)
 194                 return -1;
 195
 196         if (type != OBJ_ANY && obj->type != type)
 197                 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
 198
 199         obj->flags |= FLAG_LINK;
 200         return 0;
 201 }
 202
 203 /* The content of each linked object must have been checked
 204    or it must be already present in the object database */
 205 static unsigned check_object(struct object *obj)
 206 {
 207         if (!obj)
 208                 return 0;
 209
 210         if (!(obj->flags & FLAG_LINK))
 211                 return 0;
 212
 213         if (!(obj->flags & FLAG_CHECKED)) {
 214                 unsigned long size;
 215                 int type = sha1_object_info(obj->sha1, &size);
 216                 if (type <= 0)
 217                         die(_("did not receive expected object %s"),
 218                               sha1_to_hex(obj->sha1));
 219                 if (type != obj->type)
 220                         die(_("object %s: expected type %s, found %s"),
 221                             sha1_to_hex(obj->sha1),
 222                             typename(obj->type), typename(type));
 223                 obj->flags |= FLAG_CHECKED;
 224                 return 1;
 225         }
 226
 227         return 0;
 228 }
 229
 230 static unsigned check_objects(void)
 231 {
 232         unsigned i, max, foreign_nr = 0;
 233
 234         max = get_max_object_index();
 235         for (i = 0; i < max; i++)
 236                 foreign_nr += check_object(get_indexed_object(i));
 237         return foreign_nr;
 238 }
 239
 240
 241 /* Discard current buffer used content. */
 242 static void flush(void)
 243 {
 244         if (input_offset) {
 245                 if (output_fd >= 0)
 246                         write_or_die(output_fd, input_buffer, input_offset);
 247                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 248                 memmove(input_buffer, input_buffer + input_offset, input_len);
 249                 input_offset = 0;
 250         }
 251 }
 252
 253 /*
 254  * Make sure at least "min" bytes are available in the buffer, and
 255  * return the pointer to the buffer.
 256  */
 257 static void *fill(int min)
 258 {
 259         if (min <= input_len)
 260                 return input_buffer + input_offset;
 261         if (min > sizeof(input_buffer))
 262                 die(Q_("cannot fill %d byte",
 263                        "cannot fill %d bytes",
 264                        min),
 265                     min);
 266         flush();
 267         do {
 268                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 269                                 sizeof(input_buffer) - input_len);
 270                 if (ret <= 0) {
 271                         if (!ret)
 272                                 die(_("early EOF"));
 273                         die_errno(_("read error on input"));
 274                 }
 275                 input_len += ret;
 276                 if (from_stdin)
 277                         display_throughput(progress, consumed_bytes + input_len);
 278         } while (input_len < min);
 279         return input_buffer;
 280 }
 281
 282 static void use(int bytes)
 283 {
 284         if (bytes > input_len)
 285                 die(_("used more bytes than were available"));
 286         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 287         input_len -= bytes;
 288         input_offset += bytes;
 289
 290         /* make sure off_t is sufficiently large not to wrap */
 291         if (signed_add_overflows(consumed_bytes, bytes))
 292                 die(_("pack too large for current definition of off_t"));
 293         consumed_bytes += bytes;
 294 }
 295
 296 static const char *open_pack_file(const char *pack_name)
 297 {
 298         if (from_stdin) {
 299                 input_fd = 0;
 300                 if (!pack_name) {
 301                         static char tmp_file[PATH_MAX];
 302                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 303                                                 "pack/tmp_pack_XXXXXX");
 304                         pack_name = xstrdup(tmp_file);
 305                 } else
 306                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 307                 if (output_fd < 0)
 308                         die_errno(_("unable to create '%s'"), pack_name);
 309                 nothread_data.pack_fd = output_fd;
 310         } else {
 311                 input_fd = open(pack_name, O_RDONLY);
 312                 if (input_fd < 0)
 313                         die_errno(_("cannot open packfile '%s'"), pack_name);
 314                 output_fd = -1;
 315                 nothread_data.pack_fd = input_fd;
 316         }
 317         git_SHA1_Init(&input_ctx);
 318         return pack_name;
 319 }
 320
 321 static void parse_pack_header(void)
 322 {
 323         struct pack_header *hdr = fill(sizeof(struct pack_header));
 324
 325         /* Header consistency check */
 326         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 327                 die(_("pack signature mismatch"));
 328         if (!pack_version_ok(hdr->hdr_version))
 329                 die(_("pack version %"PRIu32" unsupported"),
 330                         ntohl(hdr->hdr_version));
 331
 332         nr_objects = ntohl(hdr->hdr_entries);
 333         use(sizeof(struct pack_header));
 334 }
 335
 336 static NORETURN void bad_object(unsigned long offset, const char *format,
 337                        ...) __attribute__((format (printf, 2, 3)));
 338
 339 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 340 {
 341         va_list params;
 342         char buf[1024];
 343
 344         va_start(params, format);
 345         vsnprintf(buf, sizeof(buf), format, params);
 346         va_end(params);
 347         die(_("pack has bad object at offset %lu: %s"), offset, buf);
 348 }
 349
 350 static inline struct thread_local *get_thread_data(void)
 351 {
 352 #ifndef NO_PTHREADS
 353         if (threads_active)
 354                 return pthread_getspecific(key);
 355         assert(!threads_active &&
 356                "This should only be reached when all threads are gone");
 357 #endif
 358         return &nothread_data;
 359 }
 360
 361 #ifndef NO_PTHREADS
 362 static void set_thread_data(struct thread_local *data)
 363 {
 364         if (threads_active)
 365                 pthread_setspecific(key, data);
 366 }
 367 #endif
 368
 369 static struct base_data *alloc_base_data(void)
 370 {
 371         struct base_data *base = xcalloc(1, sizeof(struct base_data));
 372         base->ref_last = -1;
 373         base->ofs_last = -1;
 374         return base;
 375 }
 376
 377 static void free_base_data(struct base_data *c)
 378 {
 379         if (c->data) {
 380                 free(c->data);
 381                 c->data = NULL;
 382                 get_thread_data()->base_cache_used -= c->size;
 383         }
 384 }
 385
 386 static void prune_base_data(struct base_data *retain)
 387 {
 388         struct base_data *b;
 389         struct thread_local *data = get_thread_data();
 390         for (b = data->base_cache;
 391              data->base_cache_used > delta_base_cache_limit && b;
 392              b = b->child) {
 393                 if (b->data && b != retain)
 394                         free_base_data(b);
 395         }
 396 }
 397
 398 static void link_base_data(struct base_data *base, struct base_data *c)
 399 {
 400         if (base)
 401                 base->child = c;
 402         else
 403                 get_thread_data()->base_cache = c;
 404
 405         c->base = base;
 406         c->child = NULL;
 407         if (c->data)
 408                 get_thread_data()->base_cache_used += c->size;
 409         prune_base_data(c);
 410 }
 411
 412 static void unlink_base_data(struct base_data *c)
 413 {
 414         struct base_data *base = c->base;
 415         if (base)
 416                 base->child = NULL;
 417         else
 418                 get_thread_data()->base_cache = NULL;
 419         free_base_data(c);
 420 }
 421
 422 static int is_delta_type(enum object_type type)
 423 {
 424         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 425 }
 426
 427 static void *unpack_entry_data(unsigned long offset, unsigned long size,
 428                                enum object_type type, unsigned char *sha1)
 429 {
 430         static char fixed_buf[8192];
 431         int status;
 432         git_zstream stream;
 433         void *buf;
 434         git_SHA_CTX c;
 435         char hdr[32];
 436         int hdrlen;
 437
 438         if (!is_delta_type(type)) {
 439                 hdrlen = sprintf(hdr, "%s %lu", typename(type), size) + 1;
 440                 git_SHA1_Init(&c);
 441                 git_SHA1_Update(&c, hdr, hdrlen);
 442         } else
 443                 sha1 = NULL;
 444         if (type == OBJ_BLOB && size > big_file_threshold)
 445                 buf = fixed_buf;
 446         else
 447                 buf = xmalloc(size);
 448
 449         memset(&stream, 0, sizeof(stream));
 450         git_inflate_init(&stream);
 451         stream.next_out = buf;
 452         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 453
 454         do {
 455                 unsigned char *last_out = stream.next_out;
 456                 stream.next_in = fill(1);
 457                 stream.avail_in = input_len;
 458                 status = git_inflate(&stream, 0);
 459                 use(input_len - stream.avail_in);
 460                 if (sha1)
 461                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 462                 if (buf == fixed_buf) {
 463                         stream.next_out = buf;
 464                         stream.avail_out = sizeof(fixed_buf);
 465                 }
 466         } while (status == Z_OK);
 467         if (stream.total_out != size || status != Z_STREAM_END)
 468                 bad_object(offset, _("inflate returned %d"), status);
 469         git_inflate_end(&stream);
 470         if (sha1)
 471                 git_SHA1_Final(sha1, &c);
 472         return buf == fixed_buf ? NULL : buf;
 473 }
 474
 475 static void *unpack_raw_entry(struct object_entry *obj,
 476                               union delta_base *delta_base,
 477                               unsigned char *sha1)
 478 {
 479         unsigned char *p;
 480         unsigned long size, c;
 481         off_t base_offset;
 482         unsigned shift;
 483         void *data;
 484
 485         obj->idx.offset = consumed_bytes;
 486         input_crc32 = crc32(0, NULL, 0);
 487
 488         p = fill(1);
 489         c = *p;
 490         use(1);
 491         obj->type = (c >> 4) & 7;
 492         size = (c & 15);
 493         shift = 4;
 494         while (c & 0x80) {
 495                 p = fill(1);
 496                 c = *p;
 497                 use(1);
 498                 size += (c & 0x7f) << shift;
 499                 shift += 7;
 500         }
 501         obj->size = size;
 502
 503         switch (obj->type) {
 504         case OBJ_REF_DELTA:
 505                 hashcpy(delta_base->sha1, fill(20));
 506                 use(20);
 507                 break;
 508         case OBJ_OFS_DELTA:
 509                 memset(delta_base, 0, sizeof(*delta_base));
 510                 p = fill(1);
 511                 c = *p;
 512                 use(1);
 513                 base_offset = c & 127;
 514                 while (c & 128) {
 515                         base_offset += 1;
 516                         if (!base_offset || MSB(base_offset, 7))
 517                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 518                         p = fill(1);
 519                         c = *p;
 520                         use(1);
 521                         base_offset = (base_offset << 7) + (c & 127);
 522                 }
 523                 delta_base->offset = obj->idx.offset - base_offset;
 524                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 525                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 526                 break;
 527         case OBJ_COMMIT:
 528         case OBJ_TREE:
 529         case OBJ_BLOB:
 530         case OBJ_TAG:
 531                 break;
 532         default:
 533                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 534         }
 535         obj->hdr_size = consumed_bytes - obj->idx.offset;
 536
 537         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 538         obj->idx.crc32 = input_crc32;
 539         return data;
 540 }
 541
 542 static void *unpack_data(struct object_entry *obj,
 543                          int (*consume)(const unsigned char *, unsigned long, void *),
 544                          void *cb_data)
 545 {
 546         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 547         unsigned long len = obj[1].idx.offset - from;
 548         unsigned char *data, *inbuf;
 549         git_zstream stream;
 550         int status;
 551
 552         data = xmalloc(consume ? 64*1024 : obj->size);
 553         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 554
 555         memset(&stream, 0, sizeof(stream));
 556         git_inflate_init(&stream);
 557         stream.next_out = data;
 558         stream.avail_out = consume ? 64*1024 : obj->size;
 559
 560         do {
 561                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 562                 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
 563                 if (n < 0)
 564                         die_errno(_("cannot pread pack file"));
 565                 if (!n)
 566                         die(Q_("premature end of pack file, %lu byte missing",
 567                                "premature end of pack file, %lu bytes missing",
 568                                len),
 569                             len);
 570                 from += n;
 571                 len -= n;
 572                 stream.next_in = inbuf;
 573                 stream.avail_in = n;
 574                 if (!consume)
 575                         status = git_inflate(&stream, 0);
 576                 else {
 577                         do {
 578                                 status = git_inflate(&stream, 0);
 579                                 if (consume(data, stream.next_out - data, cb_data)) {
 580                                         free(inbuf);
 581                                         free(data);
 582                                         return NULL;
 583                                 }
 584                                 stream.next_out = data;
 585                                 stream.avail_out = 64*1024;
 586                         } while (status == Z_OK && stream.avail_in);
 587                 }
 588         } while (len && status == Z_OK && !stream.avail_in);
 589
 590         /* This has been inflated OK when first encountered, so... */
 591         if (status != Z_STREAM_END || stream.total_out != obj->size)
 592                 die(_("serious inflate inconsistency"));
 593
 594         git_inflate_end(&stream);
 595         free(inbuf);
 596         if (consume) {
 597                 free(data);
 598                 data = NULL;
 599         }
 600         return data;
 601 }
 602
 603 static void *get_data_from_pack(struct object_entry *obj)
 604 {
 605         return unpack_data(obj, NULL, NULL);
 606 }
 607
 608 static int compare_delta_bases(const union delta_base *base1,
 609                                const union delta_base *base2,
 610                                enum object_type type1,
 611                                enum object_type type2)
 612 {
 613         int cmp = type1 - type2;
 614         if (cmp)
 615                 return cmp;
 616         return memcmp(base1, base2, UNION_BASE_SZ);
 617 }
 618
 619 static int find_delta(const union delta_base *base, enum object_type type)
 620 {
 621         int first = 0, last = nr_deltas;
 622
 623         while (first < last) {
 624                 int next = (first + last) / 2;
 625                 struct delta_entry *delta = &deltas[next];
 626                 int cmp;
 627
 628                 cmp = compare_delta_bases(base, &delta->base,
 629                                           type, objects[delta->obj_no].type);
 630                 if (!cmp)
 631                         return next;
 632                 if (cmp < 0) {
 633                         last = next;
 634                         continue;
 635                 }
 636                 first = next+1;
 637         }
 638         return -first-1;
 639 }
 640
 641 static void find_delta_children(const union delta_base *base,
 642                                 int *first_index, int *last_index,
 643                                 enum object_type type)
 644 {
 645         int first = find_delta(base, type);
 646         int last = first;
 647         int end = nr_deltas - 1;
 648
 649         if (first < 0) {
 650                 *first_index = 0;
 651                 *last_index = -1;
 652                 return;
 653         }
 654         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 655                 --first;
 656         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 657                 ++last;
 658         *first_index = first;
 659         *last_index = last;
 660 }
 661
 662 struct compare_data {
 663         struct object_entry *entry;
 664         struct git_istream *st;
 665         unsigned char *buf;
 666         unsigned long buf_size;
 667 };
 668
 669 static int compare_objects(const unsigned char *buf, unsigned long size,
 670                            void *cb_data)
 671 {
 672         struct compare_data *data = cb_data;
 673
 674         if (data->buf_size < size) {
 675                 free(data->buf);
 676                 data->buf = xmalloc(size);
 677                 data->buf_size = size;
 678         }
 679
 680         while (size) {
 681                 ssize_t len = read_istream(data->st, data->buf, size);
 682                 if (len == 0)
 683                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 684                             sha1_to_hex(data->entry->idx.sha1));
 685                 if (len < 0)
 686                         die(_("unable to read %s"),
 687                             sha1_to_hex(data->entry->idx.sha1));
 688                 if (memcmp(buf, data->buf, len))
 689                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 690                             sha1_to_hex(data->entry->idx.sha1));
 691                 size -= len;
 692                 buf += len;
 693         }
 694         return 0;
 695 }
 696
 697 static int check_collison(struct object_entry *entry)
 698 {
 699         struct compare_data data;
 700         enum object_type type;
 701         unsigned long size;
 702
 703         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 704                 return -1;
 705
 706         memset(&data, 0, sizeof(data));
 707         data.entry = entry;
 708         data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
 709         if (!data.st)
 710                 return -1;
 711         if (size != entry->size || type != entry->type)
 712                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 713                     sha1_to_hex(entry->idx.sha1));
 714         unpack_data(entry, compare_objects, &data);
 715         close_istream(data.st);
 716         free(data.buf);
 717         return 0;
 718 }
 719
 720 static void sha1_object(const void *data, struct object_entry *obj_entry,
 721                         unsigned long size, enum object_type type,
 722                         const unsigned char *sha1)
 723 {
 724         void *new_data = NULL;
 725         int collision_test_needed;
 726
 727         assert(data || obj_entry);
 728
 729         read_lock();
 730         collision_test_needed = has_sha1_file(sha1);
 731         read_unlock();
 732
 733         if (collision_test_needed && !data) {
 734                 read_lock();
 735                 if (!check_collison(obj_entry))
 736                         collision_test_needed = 0;
 737                 read_unlock();
 738         }
 739         if (collision_test_needed) {
 740                 void *has_data;
 741                 enum object_type has_type;
 742                 unsigned long has_size;
 743                 read_lock();
 744                 has_type = sha1_object_info(sha1, &has_size);
 745                 if (has_type != type || has_size != size)
 746                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 747                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 748                 read_unlock();
 749                 if (!data)
 750                         data = new_data = get_data_from_pack(obj_entry);
 751                 if (!has_data)
 752                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 753                 if (size != has_size || type != has_type ||
 754                     memcmp(data, has_data, size) != 0)
 755                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 756                 free(has_data);
 757         }
 758
 759         if (strict) {
 760                 read_lock();
 761                 if (type == OBJ_BLOB) {
 762                         struct blob *blob = lookup_blob(sha1);
 763                         if (blob)
 764                                 blob->object.flags |= FLAG_CHECKED;
 765                         else
 766                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 767                 } else {
 768                         struct object *obj;
 769                         int eaten;
 770                         void *buf = (void *) data;
 771
 772                         assert(data && "data can only be NULL for large _blobs_");
 773
 774                         /*
 775                          * we do not need to free the memory here, as the
 776                          * buf is deleted by the caller.
 777                          */
 778                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 779                         if (!obj)
 780                                 die(_("invalid %s"), typename(type));
 781                         if (do_fsck_object &&
 782                             fsck_object(obj, buf, size, 1,
 783                                     fsck_error_function))
 784                                 die(_("Error in object"));
 785                         if (fsck_walk(obj, mark_link, NULL))
 786                                 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
 787
 788                         if (obj->type == OBJ_TREE) {
 789                                 struct tree *item = (struct tree *) obj;
 790                                 item->buffer = NULL;
 791                                 obj->parsed = 0;
 792                         }
 793                         if (obj->type == OBJ_COMMIT) {
 794                                 struct commit *commit = (struct commit *) obj;
 795                                 if (detach_commit_buffer(commit, NULL) != data)
 796                                         die("BUG: parse_object_buffer transmogrified our buffer");
 797                         }
 798                         obj->flags |= FLAG_CHECKED;
 799                 }
 800                 read_unlock();
 801         }
 802
 803         free(new_data);
 804 }
 805
 806 /*
 807  * This function is part of find_unresolved_deltas(). There are two
 808  * walkers going in the opposite ways.
 809  *
 810  * The first one in find_unresolved_deltas() traverses down from
 811  * parent node to children, deflating nodes along the way. However,
 812  * memory for deflated nodes is limited by delta_base_cache_limit, so
 813  * at some point parent node's deflated content may be freed.
 814  *
 815  * The second walker is this function, which goes from current node up
 816  * to top parent if necessary to deflate the node. In normal
 817  * situation, its parent node would be already deflated, so it just
 818  * needs to apply delta.
 819  *
 820  * In the worst case scenario, parent node is no longer deflated because
 821  * we're running out of delta_base_cache_limit; we need to re-deflate
 822  * parents, possibly up to the top base.
 823  *
 824  * All deflated objects here are subject to be freed if we exceed
 825  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 826  * just need to make sure the last node is not freed.
 827  */
 828 static void *get_base_data(struct base_data *c)
 829 {
 830         if (!c->data) {
 831                 struct object_entry *obj = c->obj;
 832                 struct base_data **delta = NULL;
 833                 int delta_nr = 0, delta_alloc = 0;
 834
 835                 while (is_delta_type(c->obj->type) && !c->data) {
 836                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 837                         delta[delta_nr++] = c;
 838                         c = c->base;
 839                 }
 840                 if (!delta_nr) {
 841                         c->data = get_data_from_pack(obj);
 842                         c->size = obj->size;
 843                         get_thread_data()->base_cache_used += c->size;
 844                         prune_base_data(c);
 845                 }
 846                 for (; delta_nr > 0; delta_nr--) {
 847                         void *base, *raw;
 848                         c = delta[delta_nr - 1];
 849                         obj = c->obj;
 850                         base = get_base_data(c->base);
 851                         raw = get_data_from_pack(obj);
 852                         c->data = patch_delta(
 853                                 base, c->base->size,
 854                                 raw, obj->size,
 855                                 &c->size);
 856                         free(raw);
 857                         if (!c->data)
 858                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 859                         get_thread_data()->base_cache_used += c->size;
 860                         prune_base_data(c);
 861                 }
 862                 free(delta);
 863         }
 864         return c->data;
 865 }
 866
 867 static void resolve_delta(struct object_entry *delta_obj,
 868                           struct base_data *base, struct base_data *result)
 869 {
 870         void *base_data, *delta_data;
 871
 872         if (show_stat) {
 873                 delta_obj->delta_depth = base->obj->delta_depth + 1;
 874                 deepest_delta_lock();
 875                 if (deepest_delta < delta_obj->delta_depth)
 876                         deepest_delta = delta_obj->delta_depth;
 877                 deepest_delta_unlock();
 878         }
 879         delta_obj->base_object_no = base->obj - objects;
 880         delta_data = get_data_from_pack(delta_obj);
 881         base_data = get_base_data(base);
 882         result->obj = delta_obj;
 883         result->data = patch_delta(base_data, base->size,
 884                                    delta_data, delta_obj->size, &result->size);
 885         free(delta_data);
 886         if (!result->data)
 887                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 888         hash_sha1_file(result->data, result->size,
 889                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 890         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 891                     delta_obj->idx.sha1);
 892         counter_lock();
 893         nr_resolved_deltas++;
 894         counter_unlock();
 895 }
 896
 897 /*
 898  * Standard boolean compare-and-swap: atomically check whether "*type" is
 899  * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
 900  * and return false.
 901  */
 902 static int compare_and_swap_type(enum object_type *type,
 903                                  enum object_type want,
 904                                  enum object_type set)
 905 {
 906         enum object_type old;
 907
 908         type_cas_lock();
 909         old = *type;
 910         if (old == want)
 911                 *type = set;
 912         type_cas_unlock();
 913
 914         return old == want;
 915 }
 916
 917 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 918                                                   struct base_data *prev_base)
 919 {
 920         if (base->ref_last == -1 && base->ofs_last == -1) {
 921                 union delta_base base_spec;
 922
 923                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 924                 find_delta_children(&base_spec,
 925                                     &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
 926
 927                 memset(&base_spec, 0, sizeof(base_spec));
 928                 base_spec.offset = base->obj->idx.offset;
 929                 find_delta_children(&base_spec,
 930                                     &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
 931
 932                 if (base->ref_last == -1 && base->ofs_last == -1) {
 933                         free(base->data);
 934                         return NULL;
 935                 }
 936
 937                 link_base_data(prev_base, base);
 938         }
 939
 940         if (base->ref_first <= base->ref_last) {
 941                 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
 942                 struct base_data *result = alloc_base_data();
 943
 944                 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
 945                                            base->obj->real_type))
 946                         die("BUG: child->real_type != OBJ_REF_DELTA");
 947
 948                 resolve_delta(child, base, result);
 949                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
 950                         free_base_data(base);
 951
 952                 base->ref_first++;
 953                 return result;
 954         }
 955
 956         if (base->ofs_first <= base->ofs_last) {
 957                 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
 958                 struct base_data *result = alloc_base_data();
 959
 960                 assert(child->real_type == OBJ_OFS_DELTA);
 961                 child->real_type = base->obj->real_type;
 962                 resolve_delta(child, base, result);
 963                 if (base->ofs_first == base->ofs_last)
 964                         free_base_data(base);
 965
 966                 base->ofs_first++;
 967                 return result;
 968         }
 969
 970         unlink_base_data(base);
 971         return NULL;
 972 }
 973
 974 static void find_unresolved_deltas(struct base_data *base)
 975 {
 976         struct base_data *new_base, *prev_base = NULL;
 977         for (;;) {
 978                 new_base = find_unresolved_deltas_1(base, prev_base);
 979
 980                 if (new_base) {
 981                         prev_base = base;
 982                         base = new_base;
 983                 } else {
 984                         free(base);
 985                         base = prev_base;
 986                         if (!base)
 987                                 return;
 988                         prev_base = base->base;
 989                 }
 990         }
 991 }
 992
 993 static int compare_delta_entry(const void *a, const void *b)
 994 {
 995         const struct delta_entry *delta_a = a;
 996         const struct delta_entry *delta_b = b;
 997
 998         /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
 999         return compare_delta_bases(&delta_a->base, &delta_b->base,
1000                                    objects[delta_a->obj_no].type,
1001                                    objects[delta_b->obj_no].type);
1002 }
1003
1004 static void resolve_base(struct object_entry *obj)
1005 {
1006         struct base_data *base_obj = alloc_base_data();
1007         base_obj->obj = obj;
1008         base_obj->data = NULL;
1009         find_unresolved_deltas(base_obj);
1010 }
1011
1012 #ifndef NO_PTHREADS
1013 static void *threaded_second_pass(void *data)
1014 {
1015         set_thread_data(data);
1016         for (;;) {
1017                 int i;
1018                 counter_lock();
1019                 display_progress(progress, nr_resolved_deltas);
1020                 counter_unlock();
1021                 work_lock();
1022                 while (nr_dispatched < nr_objects &&
1023                        is_delta_type(objects[nr_dispatched].type))
1024                         nr_dispatched++;
1025                 if (nr_dispatched >= nr_objects) {
1026                         work_unlock();
1027                         break;
1028                 }
1029                 i = nr_dispatched++;
1030                 work_unlock();
1031
1032                 resolve_base(&objects[i]);
1033         }
1034         return NULL;
1035 }
1036 #endif
1037
1038 /*
1039  * First pass:
1040  * - find locations of all objects;
1041  * - calculate SHA1 of all non-delta objects;
1042  * - remember base (SHA1 or offset) for all deltas.
1043  */
1044 static void parse_pack_objects(unsigned char *sha1)
1045 {
1046         int i, nr_delays = 0;
1047         struct delta_entry *delta = deltas;
1048         struct stat st;
1049
1050         if (verbose)
1051                 progress = start_progress(
1052                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1053                                 nr_objects);
1054         for (i = 0; i < nr_objects; i++) {
1055                 struct object_entry *obj = &objects[i];
1056                 void *data = unpack_raw_entry(obj, &delta->base, obj->idx.sha1);
1057                 obj->real_type = obj->type;
1058                 if (is_delta_type(obj->type)) {
1059                         nr_deltas++;
1060                         delta->obj_no = i;
1061                         delta++;
1062                 } else if (!data) {
1063                         /* large blobs, check later */
1064                         obj->real_type = OBJ_BAD;
1065                         nr_delays++;
1066                 } else
1067                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
1068                 free(data);
1069                 display_progress(progress, i+1);
1070         }
1071         objects[i].idx.offset = consumed_bytes;
1072         stop_progress(&progress);
1073
1074         /* Check pack integrity */
1075         flush();
1076         git_SHA1_Final(sha1, &input_ctx);
1077         if (hashcmp(fill(20), sha1))
1078                 die(_("pack is corrupted (SHA1 mismatch)"));
1079         use(20);
1080
1081         /* If input_fd is a file, we should have reached its end now. */
1082         if (fstat(input_fd, &st))
1083                 die_errno(_("cannot fstat packfile"));
1084         if (S_ISREG(st.st_mode) &&
1085                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1086                 die(_("pack has junk at the end"));
1087
1088         for (i = 0; i < nr_objects; i++) {
1089                 struct object_entry *obj = &objects[i];
1090                 if (obj->real_type != OBJ_BAD)
1091                         continue;
1092                 obj->real_type = obj->type;
1093                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1094                 nr_delays--;
1095         }
1096         if (nr_delays)
1097                 die(_("confusion beyond insanity in parse_pack_objects()"));
1098 }
1099
1100 /*
1101  * Second pass:
1102  * - for all non-delta objects, look if it is used as a base for
1103  *   deltas;
1104  * - if used as a base, uncompress the object and apply all deltas,
1105  *   recursively checking if the resulting object is used as a base
1106  *   for some more deltas.
1107  */
1108 static void resolve_deltas(void)
1109 {
1110         int i;
1111
1112         if (!nr_deltas)
1113                 return;
1114
1115         /* Sort deltas by base SHA1/offset for fast searching */
1116         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
1117               compare_delta_entry);
1118
1119         if (verbose)
1120                 progress = start_progress(_("Resolving deltas"), nr_deltas);
1121
1122 #ifndef NO_PTHREADS
1123         nr_dispatched = 0;
1124         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1125                 init_thread();
1126                 for (i = 0; i < nr_threads; i++) {
1127                         int ret = pthread_create(&thread_data[i].thread, NULL,
1128                                                  threaded_second_pass, thread_data + i);
1129                         if (ret)
1130                                 die(_("unable to create thread: %s"),
1131                                     strerror(ret));
1132                 }
1133                 for (i = 0; i < nr_threads; i++)
1134                         pthread_join(thread_data[i].thread, NULL);
1135                 cleanup_thread();
1136                 return;
1137         }
1138 #endif
1139
1140         for (i = 0; i < nr_objects; i++) {
1141                 struct object_entry *obj = &objects[i];
1142
1143                 if (is_delta_type(obj->type))
1144                         continue;
1145                 resolve_base(obj);
1146                 display_progress(progress, nr_resolved_deltas);
1147         }
1148 }
1149
1150 /*
1151  * Third pass:
1152  * - append objects to convert thin pack to full pack if required
1153  * - write the final 20-byte SHA-1
1154  */
1155 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved);
1156 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1157 {
1158         if (nr_deltas == nr_resolved_deltas) {
1159                 stop_progress(&progress);
1160                 /* Flush remaining pack final 20-byte SHA1. */
1161                 flush();
1162                 return;
1163         }
1164
1165         if (fix_thin_pack) {
1166                 struct sha1file *f;
1167                 unsigned char read_sha1[20], tail_sha1[20];
1168                 struct strbuf msg = STRBUF_INIT;
1169                 int nr_unresolved = nr_deltas - nr_resolved_deltas;
1170                 int nr_objects_initial = nr_objects;
1171                 if (nr_unresolved <= 0)
1172                         die(_("confusion beyond insanity"));
1173                 objects = xrealloc(objects,
1174                                    (nr_objects + nr_unresolved + 1)
1175                                    * sizeof(*objects));
1176                 memset(objects + nr_objects + 1, 0,
1177                        nr_unresolved * sizeof(*objects));
1178                 f = sha1fd(output_fd, curr_pack);
1179                 fix_unresolved_deltas(f, nr_unresolved);
1180                 strbuf_addf(&msg, _("completed with %d local objects"),
1181                             nr_objects - nr_objects_initial);
1182                 stop_progress_msg(&progress, msg.buf);
1183                 strbuf_release(&msg);
1184                 sha1close(f, tail_sha1, 0);
1185                 hashcpy(read_sha1, pack_sha1);
1186                 fixup_pack_header_footer(output_fd, pack_sha1,
1187                                          curr_pack, nr_objects,
1188                                          read_sha1, consumed_bytes-20);
1189                 if (hashcmp(read_sha1, tail_sha1) != 0)
1190                         die(_("Unexpected tail checksum for %s "
1191                               "(disk corruption?)"), curr_pack);
1192         }
1193         if (nr_deltas != nr_resolved_deltas)
1194                 die(Q_("pack has %d unresolved delta",
1195                        "pack has %d unresolved deltas",
1196                        nr_deltas - nr_resolved_deltas),
1197                     nr_deltas - nr_resolved_deltas);
1198 }
1199
1200 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1201 {
1202         git_zstream stream;
1203         int status;
1204         unsigned char outbuf[4096];
1205
1206         memset(&stream, 0, sizeof(stream));
1207         git_deflate_init(&stream, zlib_compression_level);
1208         stream.next_in = in;
1209         stream.avail_in = size;
1210
1211         do {
1212                 stream.next_out = outbuf;
1213                 stream.avail_out = sizeof(outbuf);
1214                 status = git_deflate(&stream, Z_FINISH);
1215                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1216         } while (status == Z_OK);
1217
1218         if (status != Z_STREAM_END)
1219                 die(_("unable to deflate appended object (%d)"), status);
1220         size = stream.total_out;
1221         git_deflate_end(&stream);
1222         return size;
1223 }
1224
1225 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1226                                const unsigned char *sha1, void *buf,
1227                                unsigned long size, enum object_type type)
1228 {
1229         struct object_entry *obj = &objects[nr_objects++];
1230         unsigned char header[10];
1231         unsigned long s = size;
1232         int n = 0;
1233         unsigned char c = (type << 4) | (s & 15);
1234         s >>= 4;
1235         while (s) {
1236                 header[n++] = c | 0x80;
1237                 c = s & 0x7f;
1238                 s >>= 7;
1239         }
1240         header[n++] = c;
1241         crc32_begin(f);
1242         sha1write(f, header, n);
1243         obj[0].size = size;
1244         obj[0].hdr_size = n;
1245         obj[0].type = type;
1246         obj[0].real_type = type;
1247         obj[1].idx.offset = obj[0].idx.offset + n;
1248         obj[1].idx.offset += write_compressed(f, buf, size);
1249         obj[0].idx.crc32 = crc32_end(f);
1250         sha1flush(f);
1251         hashcpy(obj->idx.sha1, sha1);
1252         return obj;
1253 }
1254
1255 static int delta_pos_compare(const void *_a, const void *_b)
1256 {
1257         struct delta_entry *a = *(struct delta_entry **)_a;
1258         struct delta_entry *b = *(struct delta_entry **)_b;
1259         return a->obj_no - b->obj_no;
1260 }
1261
1262 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
1263 {
1264         struct delta_entry **sorted_by_pos;
1265         int i, n = 0;
1266
1267         /*
1268          * Since many unresolved deltas may well be themselves base objects
1269          * for more unresolved deltas, we really want to include the
1270          * smallest number of base objects that would cover as much delta
1271          * as possible by picking the
1272          * trunc deltas first, allowing for other deltas to resolve without
1273          * additional base objects.  Since most base objects are to be found
1274          * before deltas depending on them, a good heuristic is to start
1275          * resolving deltas in the same order as their position in the pack.
1276          */
1277         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
1278         for (i = 0; i < nr_deltas; i++) {
1279                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
1280                         continue;
1281                 sorted_by_pos[n++] = &deltas[i];
1282         }
1283         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
1284
1285         for (i = 0; i < n; i++) {
1286                 struct delta_entry *d = sorted_by_pos[i];
1287                 enum object_type type;
1288                 struct base_data *base_obj = alloc_base_data();
1289
1290                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1291                         continue;
1292                 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
1293                 if (!base_obj->data)
1294                         continue;
1295
1296                 if (check_sha1_signature(d->base.sha1, base_obj->data,
1297                                 base_obj->size, typename(type)))
1298                         die(_("local object %s is corrupt"), sha1_to_hex(d->base.sha1));
1299                 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
1300                                         base_obj->data, base_obj->size, type);
1301                 find_unresolved_deltas(base_obj);
1302                 display_progress(progress, nr_resolved_deltas);
1303         }
1304         free(sorted_by_pos);
1305 }
1306
1307 static void final(const char *final_pack_name, const char *curr_pack_name,
1308                   const char *final_index_name, const char *curr_index_name,
1309                   const char *keep_name, const char *keep_msg,
1310                   unsigned char *sha1)
1311 {
1312         const char *report = "pack";
1313         char name[PATH_MAX];
1314         int err;
1315
1316         if (!from_stdin) {
1317                 close(input_fd);
1318         } else {
1319                 fsync_or_die(output_fd, curr_pack_name);
1320                 err = close(output_fd);
1321                 if (err)
1322                         die_errno(_("error while closing pack file"));
1323         }
1324
1325         if (keep_msg) {
1326                 int keep_fd, keep_msg_len = strlen(keep_msg);
1327
1328                 if (!keep_name)
1329                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1330                 else
1331                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1332
1333                 if (keep_fd < 0) {
1334                         if (errno != EEXIST)
1335                                 die_errno(_("cannot write keep file '%s'"),
1336                                           keep_name ? keep_name : name);
1337                 } else {
1338                         if (keep_msg_len > 0) {
1339                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1340                                 write_or_die(keep_fd, "\n", 1);
1341                         }
1342                         if (close(keep_fd) != 0)
1343                                 die_errno(_("cannot close written keep file '%s'"),
1344                                           keep_name ? keep_name : name);
1345                         report = "keep";
1346                 }
1347         }
1348
1349         if (final_pack_name != curr_pack_name) {
1350                 if (!final_pack_name) {
1351                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1352                                  get_object_directory(), sha1_to_hex(sha1));
1353                         final_pack_name = name;
1354                 }
1355                 if (move_temp_to_file(curr_pack_name, final_pack_name))
1356                         die(_("cannot store pack file"));
1357         } else if (from_stdin)
1358                 chmod(final_pack_name, 0444);
1359
1360         if (final_index_name != curr_index_name) {
1361                 if (!final_index_name) {
1362                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1363                                  get_object_directory(), sha1_to_hex(sha1));
1364                         final_index_name = name;
1365                 }
1366                 if (move_temp_to_file(curr_index_name, final_index_name))
1367                         die(_("cannot store index file"));
1368         } else
1369                 chmod(final_index_name, 0444);
1370
1371         if (!from_stdin) {
1372                 printf("%s\n", sha1_to_hex(sha1));
1373         } else {
1374                 char buf[48];
1375                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1376                                    report, sha1_to_hex(sha1));
1377                 write_or_die(1, buf, len);
1378
1379                 /*
1380                  * Let's just mimic git-unpack-objects here and write
1381                  * the last part of the input buffer to stdout.
1382                  */
1383                 while (input_len) {
1384                         err = xwrite(1, input_buffer + input_offset, input_len);
1385                         if (err <= 0)
1386                                 break;
1387                         input_len -= err;
1388                         input_offset += err;
1389                 }
1390         }
1391 }
1392
1393 static int git_index_pack_config(const char *k, const char *v, void *cb)
1394 {
1395         struct pack_idx_option *opts = cb;
1396
1397         if (!strcmp(k, "pack.indexversion")) {
1398                 opts->version = git_config_int(k, v);
1399                 if (opts->version > 2)
1400                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1401                 return 0;
1402         }
1403         if (!strcmp(k, "pack.threads")) {
1404                 nr_threads = git_config_int(k, v);
1405                 if (nr_threads < 0)
1406                         die(_("invalid number of threads specified (%d)"),
1407                             nr_threads);
1408 #ifdef NO_PTHREADS
1409                 if (nr_threads != 1)
1410                         warning(_("no threads support, ignoring %s"), k);
1411                 nr_threads = 1;
1412 #endif
1413                 return 0;
1414         }
1415         return git_default_config(k, v, cb);
1416 }
1417
1418 static int cmp_uint32(const void *a_, const void *b_)
1419 {
1420         uint32_t a = *((uint32_t *)a_);
1421         uint32_t b = *((uint32_t *)b_);
1422
1423         return (a < b) ? -1 : (a != b);
1424 }
1425
1426 static void read_v2_anomalous_offsets(struct packed_git *p,
1427                                       struct pack_idx_option *opts)
1428 {
1429         const uint32_t *idx1, *idx2;
1430         uint32_t i;
1431
1432         /* The address of the 4-byte offset table */
1433         idx1 = (((const uint32_t *)p->index_data)
1434                 + 2 /* 8-byte header */
1435                 + 256 /* fan out */
1436                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1437                 + p->num_objects /* CRC32 table */
1438                 );
1439
1440         /* The address of the 8-byte offset table */
1441         idx2 = idx1 + p->num_objects;
1442
1443         for (i = 0; i < p->num_objects; i++) {
1444                 uint32_t off = ntohl(idx1[i]);
1445                 if (!(off & 0x80000000))
1446                         continue;
1447                 off = off & 0x7fffffff;
1448                 if (idx2[off * 2])
1449                         continue;
1450                 /*
1451                  * The real offset is ntohl(idx2[off * 2]) in high 4
1452                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1453                  * octets.  But idx2[off * 2] is Zero!!!
1454                  */
1455                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1456                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1457         }
1458
1459         if (1 < opts->anomaly_nr)
1460                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1461 }
1462
1463 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1464 {
1465         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1466
1467         if (!p)
1468                 die(_("Cannot open existing pack file '%s'"), pack_name);
1469         if (open_pack_index(p))
1470                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1471
1472         /* Read the attributes from the existing idx file */
1473         opts->version = p->index_version;
1474
1475         if (opts->version == 2)
1476                 read_v2_anomalous_offsets(p, opts);
1477
1478         /*
1479          * Get rid of the idx file as we do not need it anymore.
1480          * NEEDSWORK: extract this bit from free_pack_by_name() in
1481          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1482          * know we haven't installed this pack (hence we never have
1483          * read anything from it).
1484          */
1485         close_pack_index(p);
1486         free(p);
1487 }
1488
1489 static void show_pack_info(int stat_only)
1490 {
1491         int i, baseobjects = nr_objects - nr_deltas;
1492         unsigned long *chain_histogram = NULL;
1493
1494         if (deepest_delta)
1495                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1496
1497         for (i = 0; i < nr_objects; i++) {
1498                 struct object_entry *obj = &objects[i];
1499
1500                 if (is_delta_type(obj->type))
1501                         chain_histogram[obj->delta_depth - 1]++;
1502                 if (stat_only)
1503                         continue;
1504                 printf("%s %-6s %lu %lu %"PRIuMAX,
1505                        sha1_to_hex(obj->idx.sha1),
1506                        typename(obj->real_type), obj->size,
1507                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1508                        (uintmax_t)obj->idx.offset);
1509                 if (is_delta_type(obj->type)) {
1510                         struct object_entry *bobj = &objects[obj->base_object_no];
1511                         printf(" %u %s", obj->delta_depth, sha1_to_hex(bobj->idx.sha1));
1512                 }
1513                 putchar('\n');
1514         }
1515
1516         if (baseobjects)
1517                 printf_ln(Q_("non delta: %d object",
1518                              "non delta: %d objects",
1519                              baseobjects),
1520                           baseobjects);
1521         for (i = 0; i < deepest_delta; i++) {
1522                 if (!chain_histogram[i])
1523                         continue;
1524                 printf_ln(Q_("chain length = %d: %lu object",
1525                              "chain length = %d: %lu objects",
1526                              chain_histogram[i]),
1527                           i + 1,
1528                           chain_histogram[i]);
1529         }
1530 }
1531
1532 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1533 {
1534         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1535         const char *curr_index;
1536         const char *index_name = NULL, *pack_name = NULL;
1537         const char *keep_name = NULL, *keep_msg = NULL;
1538         struct strbuf index_name_buf = STRBUF_INIT,
1539                       keep_name_buf = STRBUF_INIT;
1540         struct pack_idx_entry **idx_objects;
1541         struct pack_idx_option opts;
1542         unsigned char pack_sha1[20];
1543         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1544
1545         if (argc == 2 && !strcmp(argv[1], "-h"))
1546                 usage(index_pack_usage);
1547
1548         check_replace_refs = 0;
1549
1550         reset_pack_idx_option(&opts);
1551         git_config(git_index_pack_config, &opts);
1552         if (prefix && chdir(prefix))
1553                 die(_("Cannot come back to cwd"));
1554
1555         for (i = 1; i < argc; i++) {
1556                 const char *arg = argv[i];
1557
1558                 if (*arg == '-') {
1559                         if (!strcmp(arg, "--stdin")) {
1560                                 from_stdin = 1;
1561                         } else if (!strcmp(arg, "--fix-thin")) {
1562                                 fix_thin_pack = 1;
1563                         } else if (!strcmp(arg, "--strict")) {
1564                                 strict = 1;
1565                                 do_fsck_object = 1;
1566                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1567                                 strict = 1;
1568                                 check_self_contained_and_connected = 1;
1569                         } else if (!strcmp(arg, "--verify")) {
1570                                 verify = 1;
1571                         } else if (!strcmp(arg, "--verify-stat")) {
1572                                 verify = 1;
1573                                 show_stat = 1;
1574                         } else if (!strcmp(arg, "--verify-stat-only")) {
1575                                 verify = 1;
1576                                 show_stat = 1;
1577                                 stat_only = 1;
1578                         } else if (!strcmp(arg, "--keep")) {
1579                                 keep_msg = "";
1580                         } else if (starts_with(arg, "--keep=")) {
1581                                 keep_msg = arg + 7;
1582                         } else if (starts_with(arg, "--threads=")) {
1583                                 char *end;
1584                                 nr_threads = strtoul(arg+10, &end, 0);
1585                                 if (!arg[10] || *end || nr_threads < 0)
1586                                         usage(index_pack_usage);
1587 #ifdef NO_PTHREADS
1588                                 if (nr_threads != 1)
1589                                         warning(_("no threads support, "
1590                                                   "ignoring %s"), arg);
1591                                 nr_threads = 1;
1592 #endif
1593                         } else if (starts_with(arg, "--pack_header=")) {
1594                                 struct pack_header *hdr;
1595                                 char *c;
1596
1597                                 hdr = (struct pack_header *)input_buffer;
1598                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1599                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1600                                 if (*c != ',')
1601                                         die(_("bad %s"), arg);
1602                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1603                                 if (*c)
1604                                         die(_("bad %s"), arg);
1605                                 input_len = sizeof(*hdr);
1606                         } else if (!strcmp(arg, "-v")) {
1607                                 verbose = 1;
1608                         } else if (!strcmp(arg, "-o")) {
1609                                 if (index_name || (i+1) >= argc)
1610                                         usage(index_pack_usage);
1611                                 index_name = argv[++i];
1612                         } else if (starts_with(arg, "--index-version=")) {
1613                                 char *c;
1614                                 opts.version = strtoul(arg + 16, &c, 10);
1615                                 if (opts.version > 2)
1616                                         die(_("bad %s"), arg);
1617                                 if (*c == ',')
1618                                         opts.off32_limit = strtoul(c+1, &c, 0);
1619                                 if (*c || opts.off32_limit & 0x80000000)
1620                                         die(_("bad %s"), arg);
1621                         } else
1622                                 usage(index_pack_usage);
1623                         continue;
1624                 }
1625
1626                 if (pack_name)
1627                         usage(index_pack_usage);
1628                 pack_name = arg;
1629         }
1630
1631         if (!pack_name && !from_stdin)
1632                 usage(index_pack_usage);
1633         if (fix_thin_pack && !from_stdin)
1634                 die(_("--fix-thin cannot be used without --stdin"));
1635         if (!index_name && pack_name) {
1636                 size_t len;
1637                 if (!strip_suffix(pack_name, ".pack", &len))
1638                         die(_("packfile name '%s' does not end with '.pack'"),
1639                             pack_name);
1640                 strbuf_add(&index_name_buf, pack_name, len);
1641                 strbuf_addstr(&index_name_buf, ".idx");
1642                 index_name = index_name_buf.buf;
1643         }
1644         if (keep_msg && !keep_name && pack_name) {
1645                 size_t len;
1646                 if (!strip_suffix(pack_name, ".pack", &len))
1647                         die(_("packfile name '%s' does not end with '.pack'"),
1648                             pack_name);
1649                 strbuf_add(&keep_name_buf, pack_name, len);
1650                 strbuf_addstr(&keep_name_buf, ".idx");
1651                 keep_name = keep_name_buf.buf;
1652         }
1653         if (verify) {
1654                 if (!index_name)
1655                         die(_("--verify with no packfile name given"));
1656                 read_idx_option(&opts, index_name);
1657                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1658         }
1659         if (strict)
1660                 opts.flags |= WRITE_IDX_STRICT;
1661
1662 #ifndef NO_PTHREADS
1663         if (!nr_threads) {
1664                 nr_threads = online_cpus();
1665                 /* An experiment showed that more threads does not mean faster */
1666                 if (nr_threads > 3)
1667                         nr_threads = 3;
1668         }
1669 #endif
1670
1671         curr_pack = open_pack_file(pack_name);
1672         parse_pack_header();
1673         objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1674         deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
1675         parse_pack_objects(pack_sha1);
1676         resolve_deltas();
1677         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1678         free(deltas);
1679         if (strict)
1680                 foreign_nr = check_objects();
1681
1682         if (show_stat)
1683                 show_pack_info(stat_only);
1684
1685         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1686         for (i = 0; i < nr_objects; i++)
1687                 idx_objects[i] = &objects[i].idx;
1688         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1689         free(idx_objects);
1690
1691         if (!verify)
1692                 final(pack_name, curr_pack,
1693                       index_name, curr_index,
1694                       keep_name, keep_msg,
1695                       pack_sha1);
1696         else
1697                 close(input_fd);
1698         free(objects);
1699         strbuf_release(&index_name_buf);
1700         strbuf_release(&keep_name_buf);
1701         if (pack_name == NULL)
1702                 free((void *) curr_pack);
1703         if (index_name == NULL)
1704                 free((void *) curr_index);
1705
1706         /*
1707          * Let the caller know this pack is not self contained
1708          */
1709         if (check_self_contained_and_connected && foreign_nr)
1710                 return 1;
1711
1712         return 0;
1713 }